Preliminary System Identification of Dragonfly’s Octocopter

Abstract

The Johns Hopkins University Applied Physics Laboratory (APL) is leading Dragonfly, a mission to study the prebiotic chemistry of Titan, one of Saturn’s moons. Given Titan’s diverse surface environments, mobility is crucial to the science mission, so controls engineers are faced with the challenge of designing an autonomous flight-control system for an aerial vehicle that will operate in uncertain environments. Part of the flight controller development approach involves testing with a half-scale test vehicle in an Earth environment; and one part of this process is system identification. Here, we detail the design and testing of the first round of system identification experiments with the test vehicle in which random-phase multisines were injected into the attitude commands during hover. Four experiments were performed using the half-scale test vehicle. Because of significant wind disturbances, the collected data had low coherence and were ultimately unsuitable for nonparametric frequency response estimation. System identification is an iterative process, and we present several planned ways to improve the coherence of the flight data.